The busybox size is increased by 3.2KByte uncompressed on bcm47xx with
this commit.
This zram-swap automatically creates a zram device, a swap partition on
it and make the kernel swap pages to it.
Thank you Bastian Bittorf <bittorf@bluebottle.com> for idea and the
script.
SVN-Revision: 35025
I wanted to use some statistics from /sys/kernel/debug/ieee80211,
so I enabled PACKAGE_MAC80211_DEBUGFS. But I could not use it,
because CONFIG_KERNEL_DEBUG_FS was not set and it's a little bit
hidden. After manually setting CONFIG_KERNEL_DEBUG_FS to y,
everything was fine. This patch automatically selects
CONFIG_KERNEL_DEBUG_FS if PACKAGE_MAC80211_DEBUGFS is set.
Additionally, the help text says "Say N unless you know you need this.",
but it's default y. This help text is copied verbatim from the
kernel's Kconfig and not true for us. So remove this part of the
help text
Signed-off-by: Stefan Hellermann <stefan@the2masters.de>
SVN-Revision: 35014
This adds diag leds/buttons support for Linksys E1000 V1. It worked on
my router, however it may need some tweaking by people more familiar
with openwrt and c to make it fit for commiting.
Thank you hhm for this patch.
SVN-Revision: 34996
I wrote this patch some time ago because I had a need for one of the
Gigabit ports (Linksys E3000) to be forced to 100FD. This is based
on the robocfg sources included w/ the RT-N16 sources from ASUS.
Since work is progressing on a BGMAC driver that could be included in
OpenWRT, this may be useful to someone else.
In testing, forcing the speed to 10/100 or 1000 worked fine; however,
when trying to force full-duplex mode, the result was always
half-duplex. I was not able to isolate the source of the problem
(this patch, driver or H/W limitation). The only way I could get it
to work was to set the port to Auto, but then only advertise 100FD
(not included in this patch).
I have a modified version of the robocfg package as well, I'd have to
clean it up a little first (remove the full-duplex hack) before
submitting it if there is interest.
Signed-off-by: Nathan Hintz <nlhintz@hotmail.com>
SVN-Revision: 34992
Neither --create-backup nor --restore-backup need an image name (following
the backup filename). Treat them in uniform way.
Mostly fixes#12346.
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
SVN-Revision: 34974
- introduce "list interface" options to specify the ifaces to listen on, takes uci- or network device name
- set the default system description to distrib @ hostname, e.g. "OpenWrt Barrier Breaker r34744 @ vbox"
- introduce "option lldp_description" to override the automatically generated description
SVN-Revision: 34970
Calculate complements by using awk's xor() function with a mask of 0xffffffff
instead of relying on the compl() function which appears to produce broken
results on certain 64bit architectures.
SVN-Revision: 34875
procd is the new OpenWrt process management daemon. It keeps track of processes
started from init scripts (via ubus calls), and can suppress redundant service
start/restart requests when the config/environment has not changed.
SVN-Revision: 34865
Sysupgrade defines its very own pivot() function.
Prior merging boot.sh and functions.sh sysupgrade just included boot.sh,
now it includes functions.sh which defines pivot() as well, however
slightly different which causes sysupgrade to fail.
This is a hot-fix to unbreak sysupgrade, however those two pivot()
functions should actually get merged.
SVN-Revision: 34815
The behaviour of calling 'mount' differed depending on whether it called
the busybox-mount, the mount of util-linux, the mount defined in
/lib/functions.sh and /lib/functions/boot.sh
/etc/preinit even included /lib/functions.sh and /lib/functions/boot.sh,
both re-defining 'mount'.
SVN-Revision: 34792
The prefix / causes executables like mount, umount, etc. to be placed
in /bin /sbin which therewith replace the busybox symlinks.
mount and umount e.g. are linked again libmount.so.
This is e.g. fatal during a sysupgrade, where /bin/mount,
/bin/umount, etc. but not libmount.so get copied into a ramdisk
which will result in non working executables within the ramdisk.
SVN-Revision: 34791
- support reading inactive gateways and DNS information in
network_get_gateway(), network_get_dnsserver() and network_get_dnssearch()
by passing "true" as optional last argument
- internally cache fetched values to speed up subsequent accesses to the same
data, introduce network_flush_cache() to clear them
- add some inline function documentation
SVN-Revision: 34722
Using variables from the outer scope unnecessarily complicates the code and
leads to issues.
This patch fixes the bug when having an "adhoc" wifi-iface section before a
"sta" section prevents wpa_supplicant from using the key specified in the
corresponding section as it tries to use the "adhoc" key instead (1 by
default).
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
SVN-Revision: 34716
This was done previously when dhcp was handled by the network scripts.
So netifd should behave the same.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
SVN-Revision: 34704
Hauke Mehrtens
Felix Fietkau
John Crispin
Steven Barth
Jo-Philipp Wich
Gabor Juhos
Mirko Vogt
Luka Perkov